Search Results

Autonomous vehicles are currently a subject of great interest and there is heavy research on creating and improving algorithms for detecting objects in their vicinity. A ROS-based deep learning approach has been developed to detect objects using point cloud data. With encoded raw light detection and ranging (LiDAR) and camera data, several basic statistics such as elevation and density are generated. The system leverages a simple and fast convolutional neural network (CNN) solution for object identification and localization classification and generation of a bounding box to detect vehicles, pedestrians and cyclists was developed. The system is implemented on an Nvidia Jetson TX2 embedded computing platform, the classification and location of the objects are determined by the neural network. Coordinates and other properties of the object are published on to various ROS topics which are then serviced by visualization and data handling routines.

The ‘boundary of space’ model representing all possible positions which may be occupied by a mechanism during its normal range of motion (for all positions and orientations) is called the work envelope. In the robotic domain, it is also known as the robot operating envelope or workspace. Several researchers have investigated workspace boundaries for different degrees of freedom (DOF), joint types and kinematic structures utilizing many approaches. The work envelope provides essential boundary information, which is critical for safety and layout concerns, but the work envelope information does not by itself determine the reach feasibility of a desired configuration. The effect of orientation is not captured as well as the coupling related to operational parameters. Included in this are spatial occupancy concerns due to linking multiple kinematic chains, which is an issue with multi-tasking machine tools, and manufacturing cells.

When performing trajectory planning for robotic applications, there are many aspects to consider, such as the reach conditions, joint and end-effector velocities, accelerations and jerk conditions, etc. The reach conditions are dependent on the end-effector orientations and the robot kinematic structure. The reach condition feasibility is the first consideration to be addressed prior to optimizing a solution. The ‘functional’ work space or work window represents a region of feasible reach conditions, and is a sub-set of the work envelope. It is not intuitive to define. Consequently, 2D solution approaches are proposed. The 3D travel paths are decomposed to a 2D representation via radial projections. Forward kinematic representations are employed to define a 2D boundary curve for each desired end effector orientation.

Gantry robots are mainly employed for applications requiring large workspace, with limited higher manipulability in one direction than the others. The Gantries offer very good mechanical stiffness and constant positioning accuracy, but low dexterity. Common gantries are CNC machines with three translational joints XYZ (3DOF) and usually with an attached wrist (+3DOF). The translational joints are used to move the tool in any position in the 3D workspace. The wrist is used to orient the tool by rotation about X, Y and Z axis. This standard kinematic structure (3T3R) produces a rectangular workspace. In this paper a full kinematic model for a 6DOF general CNC (gantry) machine is presented, along with the Jacobian matrix and singularity analysis. Using Denavit-Hartenberg convention, firstly, the general kinematic structure is presented, in order to assign frames at each link. The forward kinematic problem is solved using Maple 17 software.

In this study the capabilities of a semi-active suspension and an active roll suspension are evaluated for comparison with a passive suspension. The vehicle used is a utility truck modeled as a multi-body system in ADAMS/Car while the ECU (electronic control unit) is built in Matlab/Simulink. Cosimulation is used in linking the vehicle model with the controller by exchanging the input and output values of each sub-system with one another. For the simulation models considered, results indicate that for a fish-hook cornering maneuver the semi-active suspension is limited in increasing vehicle performance while the active roll suspension significantly improves it. Further analysis is needed to confirm these findings.

The paper describes a ‘virtual motorsports’ event developed by the University of Windsor Vehicle Dynamics and Control Research Group. The event was a competitive project-based component of a Vehicle Dynamics course offered by the University's Department of Mechanical, Automotive, & Materials Engineering. The simulated race was developed to provide fourth year automotive engineering students with design and race experience, similar to that found in Formula SAE®or SAE Baja®, but within the confines of a single academic semester. The project, named ‘Formula463’, was conducted entirely within a virtual environment, and encompassed design, testing, and racing of hi-fidelity virtual vehicle models. The efficacy of the Formula463 program to provide students with a design experience using model based simulation tools and methods has been shown over the past two years. All of the software has been released under a General Public License and is freely available on the authors website.

Heavy traffic volume makes tailgating a common picture on the road today. Wake interference, particularly in the scenario when a relatively small sedan drives into the wake of a large truck, may raise some serious highway safety concerns. In this paper, the characteristics of the separation bubble of model trucks with various degrees of details are studied. The objective is to find out the impact of truck model details on the characteristics of the wake bubble. Our wind tunnel results revealed that the degree of model detail has a significant effect on the wake bubble; the bubble length increases with model details.

The purpose of this study is to determine the feasibility of simulating an active suspension using cosimulation. The vehicle used is a utility truck created in ADAMS/View while the E.C.U. (electronic control unit) is implemented in Simulink for both a fully-active and semi-active controller. The LQR (Linear Quadratic Regulator) is used for the fully-active system while the semi-active system uses a switching law adopted from Karnopp et al. {1}. Nonlinear and linear vehicle models are compared and the influence of suspension bushings is examined. All simulations undertaken are geared towards evaluating the ride capabilities of such systems.

Neural networks are flexible modeling tools that can be used in conjunction with multi-body dynamics models to better predict nonlinear behaviour of components. This paper focuses on a process that incorporates a neural network model of a nonlinear damping force into a single degree of freedom mass-spring-damper model. Software tools and their interaction are specified. The verification of this process is the focal point of this paper and is a necessary step before further correlation studies can be performed on more complex component representations.